**** BEGIN LOGGING AT Wed Mar 03 03:00:25 2021 Mar 03 10:20:22 hi Mar 03 10:20:27 lo Mar 03 10:20:28 I have a question on pinmuxing Mar 03 10:21:10 go on the, ask Mar 03 10:21:13 *then Mar 03 10:21:27 * mru points at $topic Mar 03 10:22:50 I have a device connected to my processor and certain pins of it should be configured a certain way but it's not associated with any driver. the device tree has the am33xx_pinmux node, can I assume that any child node with pinmuxing under it will be configured whether the pin is used by a driver or not? Mar 03 10:23:35 no Mar 03 10:24:18 those child nodes are possible mux settings, many of them conflicting Mar 03 10:24:29 something has to reference the ones you actually want to use Mar 03 10:24:30 so something needs to reference that child node for the pinmuxing to take effect? Mar 03 10:25:35 It's a reset line by the way and it won't be controlled by the processor by any driver that is why i was wondering that. Mar 03 10:25:42 if there's no device node corresponding to the function, you can add such a reference to the am33xx_pinmux node itself Mar 03 10:26:35 so you meen instead of making a child node i directly place the pinmuxing settings inside the am33xx_pinmux node? Mar 03 10:28:26 something like this: &am33xx_pinmux { pinctrl-names = "default"; pinctrl-0 = <&foo_pins>; } Mar 03 10:28:52 oh i see Mar 03 10:29:56 in my case there is already a pingctrl-0 setup for a clkout2 pin but i can add a reference to another node just for that one gpio Mar 03 10:30:34 right, the beaglebone already configures clkout2 that way Mar 03 10:30:49 so you can add whatever else you need Mar 03 10:33:59 thank you for these insights :) Mar 03 13:18:18 mru: say i want to hog that reset pin where do i place that bit? Mar 03 13:27:33 tried adding this: https://termbin.com/yb6g Mar 03 13:27:41 but it tels me it's a syntax error Mar 03 13:28:58 the comma is a syntax error Mar 03 13:29:27 oh i see Mar 03 13:46:50 Duality: pinmux always needs to be associated with a device Mar 03 13:48:49 attaching it to &am33xx_pinmux itself is possible though of course you need to be careful since it's not a very modular thing to do, e.g. if the base dts had the same idea you could end up accidently overwriting it Mar 03 13:49:44 I am not sure if I understand what you meen Mar 03 13:51:16 &am33xx_pinmux { pinctrl-names = "default"; pinctrl-0 = <&foo_pins>; } would overwrite any previous value of those properties if they were already defined previously, that's all I meant to point to Mar 03 13:52:33 instead of using gpio hog and putting pinmux on a random device that seemed convenient, you could also create a gpio-of-helper device to setup/initialize gpios and attach your pinmux node to that Mar 03 13:53:23 that's not in mainline linux though Mar 03 13:53:42 true, unfortunately Mar 03 13:54:44 but if you're not using an rcn kernel you can still easily patch that in yourself Mar 03 13:54:48 mru: what you refer to not being in mainline is gpio-of-helper ?\ Mar 03 13:55:00 right Mar 03 13:55:03 yeah the gpio-of-helper driver is not in mainline Mar 03 13:55:51 if you don't want to patch the kernel with that, and you're feeling slightly evil, you could declare a bogus fixed-regulator or something :) Mar 03 13:56:06 mainline has no way of setting up gpios for userspace access Mar 03 13:56:26 and the only way to setup gpios at all is the gpio-hog mechanism, which has no way of associating pinmux Mar 03 13:57:31 mainline just isn't always very in touch with the needs of embedded applications Mar 03 13:57:37 there's some sort of de facto rule that you should always create a kernel driver, never access things from userspace Mar 03 13:57:40 see also spidev Mar 03 13:57:44 yeah, fucking idiots Mar 03 13:58:00 sometimes a kernel driver is the best solution Mar 03 13:58:04 sometimes it's not Mar 03 13:58:30 usually tossing application-specific code into kernelspace when it could have run perfectly fine in userspace is overall a bad bad idea Mar 03 13:59:03 it all depends on what you're actually doing Mar 03 14:00:04 if PCs of old didn't have serial ports, we probably wouldn't be allowed /dev/ttyS either Mar 03 14:00:13 the spidev thing is particularly asinine since SPI is just a way to communicate, just like an uart or i2c or the many other ways you can communicate with devices Mar 03 14:00:28 of course both beagleboard and rpi kernels reverted that patch Mar 03 14:00:36 everybody does Mar 03 14:02:39 like I said, kernel devs not being in touch with what people actually want/need :P Mar 03 14:02:53 and honestly there's no good justification for it in case of spidev Mar 03 14:09:40 m Mar 03 14:53:20 woa, i just found a new way for a devicetree to fail. Not even uboot enters, it just shuts back down Mar 03 14:54:16 uhh, how would a DT cause that? Mar 03 14:54:32 uboot is the one loading the DT Mar 03 15:02:24 dunno, maybe uboot is loading it, but then it reboots the system. time to break out the uart Mar 03 15:11:49 so the rpi doesn't have spidev anymore ? Mar 03 15:12:33 Duality: ?? Mar 03 15:13:05 Duality: no, mainline made it difficult to use the spidev driver and both beaglebone and rpi reverted that patch to ensure spidev is usable Mar 03 15:13:21 oh Mar 03 15:13:32 I totally interpreted that wrong my bad :) Mar 03 15:13:53 Cause that would have been crazy Mar 03 15:14:11 but what did they do to spidev? Mar 03 15:16:45 print a big warning in kernel log if DT uses compatible="spidev" Mar 03 15:17:49 big warning as in WARN(), i.e. complete with stack traceback Mar 03 15:18:00 why Mar 03 15:18:05 that seems silly Mar 03 15:18:38 I once tried to troll them with this: https://lore.kernel.org/lkml/1453394169-11458-1-git-send-email-mans@mansr.com/ Mar 03 15:20:19 lol Mar 03 15:20:52 I think that actually ended up in production somewhere Mar 03 15:21:02 hahaha, of course it did Mar 03 15:21:14 unless they abandoned the product Mar 03 15:22:13 I keep clear of mobile cranes, just in case Mar 03 15:23:53 lol Mar 03 15:24:13 I think for many products we should be happy we're unaware of what's running on them Mar 03 15:24:32 oh yes Mar 03 15:25:05 I have some industrial-looking thing on my desk right now that I have no idea where it's headed Mar 03 15:50:09 Hi, are their people here? Mar 03 15:50:21 no, only trolls Mar 03 15:50:29 161 bots in fact Mar 03 15:50:34 Im wanting to build a robot on a longboard and want advice on the best cortex to use Mar 03 15:51:16 ehm Mar 03 15:51:32 need more info im assuming? Mar 03 15:52:22 that question is so vague as to not have any answer. "the best cortex" is also strangely formulated since you don't generally start with picking a cpu core, as that's probably the least interesting part of the hardware setup Mar 03 15:52:53 really? im not used to having a choice. Mar 03 15:53:14 I worked with vex robotics in high school and thats the extent of my knowledge Mar 03 15:54:44 Im wanting to make a longboard i can control and steer with my phone/controller of some sort. I want to make it eventually drive itself but i need to know programming to do that. I thought i needed to start on figuring out between like an arduino and a beagleboard before i starrted buying motors do to campatability. Mar 03 15:54:45 I mean, you'd probably start with picking a board that meets your requirements, and whatever cpu it has is whatever cpu it has Mar 03 15:55:44 the problem here is also that you're clearly quite some distance away from your final goal with many things to be learned, I don't really know how to advise you on the path to take Mar 03 15:56:55 i mean i have to start somewhere. thats why i came here. Im also in school currently for things similar to this type of stuff so i will be learning here soon. I just kinda wanted a project to work on Mar 03 15:57:51 I was going to suggest an EE degree as a good first step Mar 03 15:58:16 Ooh, ooh, ooh! Mar 03 15:58:22 sure but it sounds like you need to learn programming, if you're going to use an embedded linux system like a beaglebone you'll want some basic linux knowledge, regardless of what board you use you'll need some amount of hardware knowledge if you want to avoid a big plume of smoke arising from your setup on day 1, etc Mar 03 15:58:50 Hmm. Plumes. Mar 03 15:59:03 there can be a big plume? I must be doing something wrong then Mar 03 15:59:29 I've rarely had more than a crackle and a faint whiff Mar 03 15:59:30 Im working on be an ETN. Not the exact same as an EE but some simularities Mar 03 15:59:48 i cant spell very well, sorry Mar 03 15:59:52 an arduino is much more limited than a beaglebone, but there are a lot of resource for it, and it's cheaper and most robust to electrical abuse Mar 03 16:00:49 i dont know much about programming in linux. I know java decently but not much about anything else Mar 03 16:01:56 Get one and become a die-hard fanatic like me! Mar 03 16:02:31 Then, we can revisit ideas together, right? Mar 03 16:03:39 I tried. W/ a hammer drill in the background, it is hard to think straight right now. Sorry for interrupting. Mar 03 16:03:43 jackgrey1207: I'm not sure why you're contrasting programming on linux (an operating system) with programming in java (a programming language) .... which programming language you use and which operating system you use are pretty much completely independent matters Mar 03 16:04:06 You scared him off. Mar 03 18:53:43 can you load multiple dts overlays that overlay the &{/} path Mar 03 18:54:01 actually scratch that question. Mar 03 19:08:26 any node can be overlayed any number of times both by different overlays or even a single overlay Mar 03 19:15:07 Apparently the issue of poweroff is actually a kernel panic from the tilcdc probe...merp Mar 03 19:15:32 ...kernel panic sounds like a band name Mar 03 19:17:44 Colonel Panic and the Page Faults Mar 03 19:36:09 Is the BeagleBone AI RoHS compliant? Mar 03 19:36:51 has anyone successfully run the fbtft drivers on the BeagleAI? Mar 03 19:45:06 What's wrong with this DTS? https://pastebin.com/2A210quq Mar 03 19:46:19 its indenting is a complete mess? :P Mar 03 19:47:04 also you should not override #address-cells or #size-cells on nodes you're overlaying. the warning is bogus, silence it using dtc flags Mar 03 19:48:04 anyway, is there something specific you want people to look at? Mar 03 19:49:43 also the status="okay"; on your backlight node is silly and pointless Mar 03 19:50:42 Good evening BB community, Mar 03 19:50:42 after having had very mixed experiences with the RaspberryPi over the last few weeks, I'm looking around for an alternative for it. Mar 03 19:50:43 At the moment I'm programming the Pi in Visual Studio with the .NET Framework and the corresponding System.Device Library. Mar 03 19:50:43 This all works perfectly, however I am having problems with the PI's I2C bus. Mar 03 19:50:44 Due to the faulty clock-stretching I am unfortunately not able to write or read data to an IO-Expander CY8C9520A. Mar 03 19:50:44 Now I would like to know if someone here has made experiences with I2C and clock-stretching on the BBB and can assure me that it is no problem with the BBB. Mar 03 19:50:45 Furthermore it would be very interesting for me if one of you has ever built .NET Apps with the System.Device Library on the BBB. Mar 03 19:50:45 I would be very grateful for your help. Mar 03 19:50:47 Honestly I have no idea. Mar 03 19:51:10 cocrea: clock stretching works fine yeah Mar 03 19:51:11 been poking things around, but my kernel panic issue may be just from my kernel compile Mar 03 19:52:16 zmatt: uuh thats nice to know, thank you for the fast answer. Mar 03 19:53:10 cocrea: I know for a fact that TI's i2c controller correctly deals with clock stretching on every single clock cycle, since I've used that to single-step through i2c transactions (using a microcontroller that would stretch every clock cycle until I press a key) Mar 03 19:55:28 Konsgnx1: oh also you're creating a new device node for tilcdc (as /fb) .. don't. Mar 03 19:55:38 it already exists as &lcdc Mar 03 19:56:10 just set your desired blue-and-red-wiring on &lcdc Mar 03 19:56:18 along with status = "okay"; Mar 03 19:56:29 alright alright, does the panel also go in there? Mar 03 19:58:52 no your panel goes in / Mar 03 19:59:09 zmatt: Thanks alot. The BBB AM335 is using a TI processor with a TI i2c controller right? I ask as a precaution because I actually expected from the RPi also that clock stretching works Mar 03 19:59:40 Thank you! Mar 03 20:01:53 cocrea: yep, specifically the TI OMAP I2C controller Mar 03 20:02:17 (TI's keystone devices use a different I2C controller I think) Mar 03 20:02:36 Konsgnx1: notes on DT config for lcd: https://pastebin.com/xnPyKNfi Mar 03 20:03:49 hahah, zmatt, do you keep a book of notes on the sitara and all the aspects of it. you always have something super helpful. Mar 03 20:04:30 I've collected a fair bit of stuff over the years Mar 03 20:05:01 also please don't say "the sitara", that word is a meaningless marketing umbrella for a collection of very different SoCs Mar 03 20:05:02 would it be valid to make that root as an overlay like &{/}? Mar 03 20:05:11 gotcha Mar 03 20:05:25 yeah if used as overlay you'd need to use &{/} instead of / of course Mar 03 20:08:42 zmatt thats really good to know, thank you alot for your help :) Do you know by any chance if the standard .NET / AspNet Core Runtime especially the System.Device Library are working on the BBB ? Mar 03 20:09:23 that's not a BBB question... if they're working on linux, they're working on the BBB Mar 03 20:10:16 unless it's a library for specific hardware, in which case consult documentation for the library Mar 03 20:10:31 zmatt Okay thanks:) Mar 03 20:10:37 System.Device seems to be a namespace, not a library, based on what I'm finding with google Mar 03 20:14:03 zmatt that was my mistake. the library is called .NET iot Mar 03 20:19:29 Konsgnx1: I also have this nice pinout table for lcdc: https://goo.gl/Jkcg0w#gid=1280625524 Mar 03 20:21:12 alright, that is extremely thorough Mar 03 20:22:11 or if you need to connect to MIPI DPI, see the next sheet ;) Mar 03 20:22:42 (though I've never had to, so that sheet isn't verified particularly well) Mar 03 22:31:57 Hello...do I need the SGX lib. from ImgTec to work w/ graphics directly on a LCD screen. I know I will most likely need to figure out a ton more. Mar 03 22:32:43 But, I kind of already posted in their forum and the instructions on their Native_SDK states I need it but there was only one SGX lib. I found on their site. Mar 03 22:34:23 SGX 530 is on the BBB but I am sure there are more than just one SGX lib. b/c of the other SGX processors. Hmm. Mar 03 22:37:59 Well, anyway... Mar 03 22:38:09 I will keep locating these libs. slowly. Mar 03 23:53:47 bbl! Mar 04 01:17:20 lol, when building a shared library that's linked to some static library, the symbols imported from that static library get reexported from the shared library _by default_ ... who the hell thought that was a sane default behaviour *facepalm* **** ENDING LOGGING AT Thu Mar 04 02:59:56 2021